The aim of this study was to investigate occupational cognitive complexity of main lifetime occupation in relation to level and 15-year change in episodic memory recall in a sample of older adults (≥ 65 years, n = 780). We used latent growth curve modelling with occupational cognitive complexity (O*NET indicators) as independent variable. Subgroup analyses in a sample of middle-aged (mean: 49.9 years) men (n = 260) were additionally performed to investigate if a general cognitive ability (g) factor at age 18 was predictive of future occupational cognitive complexity and cognitive performance in midlife. For the older sample, a higher level of occupational cognitive complexity was related to a higher level of episodic recall (β = 0.15, p < .001), but the association with rate of change (β = 0.03, p = .64) was not statistically significant. In the middle-aged sample, g at age 18 was both directly (β = 0.19, p = .01) and indirectly (via years of education after age 18, ab = 0.19) predictive of midlife levels of occupational cognitive complexity. Cognitive ability at age 18 was also a direct predictor of midlife episodic recall (β = 0.60, p ≤ 0.001). Critically, entry of the early adult g factor attenuated the association between occupational complexity and cognitive level (from β = 0.21, p = .01 to β = 0.12, p = .14). Overall, our results support a pattern of preserved differentiation from early to late adulthood for individuals with different histories of occupational complexity.